A new study by UCLA researchers, published online on Nov. 14, 2014 in Cancer Imaging, features an innovative way to overcome traditional limitations in brain cancer imaging technology that have long frustrated physicians' ability to properly monitor the treatment of glioblastomas. A glioblastoma is a particularly aggressive type of brain tumor that grows from normal brain cells in the supportive tissue of the brain. The large network of surrounding blood vessels increases, a process called hypervascularity, and feeds the tumor cells as they reproduce and grow. This is a key factor in identifying the abnormal brain tissue through imaging. Chemotherapy drug treatment focuses on a class of drugs called anti-angiogenics, which aim to halt the growth of these new blood vessels and starve the tumor of nutrients. But because glioblastomas have many different types of abnormalities, patients on the same treatment respond to it differently — and patients with the same tumor can have different outcomes, which are currently unpredictable.
Looking beyond the tumor in brain cancer imaging
Traditional brain cancer imaging measures the amount of blood supplied to the tumor, called tumor perfusion. "When studying brain tumors, we are usually looking at what's tumor and what's not tumor," explains lead study author Dr. Benjamin Ellingson of the Department of Radiological Sciences at the David Geffen School of Medicine at UCLA. "But we found that just studying the blood volume in the tumor didn't seem to predict whether patients were doing well." Instead, the UCLA researchers wanted to study the entire hypervascular area around the tumor to see if it was responding to the chemotherapy by shrinking, staying the same or expanding.
For this new study, Dr. Ellingson and nine other UCLA researchers constructed a large-scale population atlas comprised of brain cancer imaging data from 450 patients chosen from the UCLA brain cancer imaging database, one of the largest in the world. They created the atlas using only the normal brain tissue, comparing it against the cerebral blood volume maps of individual glioblastoma patients. This method helped define and measure changes within the entire hypervascular area around the tumor, predicting a patient's response to the treatment more accurately.
Comparison to a large-scale population atlas improves data
The researchers theorized that differences in magnetic resonance scanners and sequence parameters cause variability in measurements, and may contribute to a lack of correlation between traditional tumor scans and tumor response to treatment. To account for these differences, the researchers created the atlas — which included tissue from a very wide range of patients under a variety of scan protocols — for comparison, to identify abnormal hypervascular tissue more distinctly in glioblastoma patients. Because normal perfusion parameters differ depending on where the tumor is in the brain, the researchers created thresholds within the atlas to account for the different tissue types, which had been hard to measure and compare with traditional perfusion imaging, explains Dr. Ellingson.
Specifically, the researchers measured the volume of hypervascularity before and after treatment, the percentage of hypervascular tumor within contrast enhancement areas and the mean cerebral blood volume of hypervascular tissue for each of the 32 scans in the analysis (64 in total).
A new way to predict gliobastoma patient response
"We were able to identify the features of the entire hypervascular area that pointed directly to the patient's response to the chemotherapy treatment in a way that measuring only the blood volume of the tumor did not," says Dr. Ellingson.
The results of studying the extent of the entire hypervascular area around the tumor by comparing individual glioblastoma patients to the normal brain tissue atlas proved predictive. The UCLA researchers plan to study different cancer types using large population atlases, and to work with other countries to create an even larger international atlas.
